Do you have a question? Post it now! No Registration Necessary. Now with pictures!
- Posted on
- Low Voltage on 12V Rail
March 7, 2010, 6:06 am
rate this thread
I've got a beat-up HP on the bench, and it appears to be functioning
normally. Windows XP startup was hosed such that it would not make it
to the desktop--always looping back to the boot options menu. A
reinstall of the OS has solved that problem. No S.M.A.R.T. errors have
been tripped, and the drive checks out in SeaTools.
The power supply, though, might be a different matter. I hooked it up
to one of those antec power supply testers, and it's worried (beeping)
about the 12V supply coming in at about 11.5. I would really like to
find "the problem" with this machine, but is that reading really a big deal?
Thank you for your insights.
Re: Low Voltage on 12V Rail
Table 2 on PDF page 13 says the main rails have a tolerance of 5%.
So +11.40 to +12.60 is the range to be expected. The power supply
designer likely expects the design to do better than that. (Because the
person who designed the supply, really doesn't want a significant percentage
of the units to be "dragging their heels" and come back for warranty reasons.)
In some cases, they dial the knob a bit on the high side, to prevent the thing
from getting too close to the low limit. A cheap supply may not be exactly
centered on +12.0V (although the spec intends that to be the case).
You should really verify with a proper multimeter. I usually find the multimeter
shows the supply to be in excellent health, while things like the hardware
monitor on the motherboard is way off. We have no idea what the tolerances are
like on the Antec item. Antec makes computer cases, and any electronics they
make, are likely not designed in house. Remember the crappy USB front ports
they made ? Clueless, when it comes to electrical things. Antec is relying
on someone else, for the intelligence of the design.
I also recommend a clamp-on DC ammeter, as that allows you to quickly check
for abnormal loading by any connected load. A supply can be low, because
the supply is failing (feedback system cranked to the wall, but output
still low). Or, a supply can be low, because a fault in a motherboard circuit
or peripheral is dragging the supply down. There was at least one case,
where the entire area around a CPU socket was charred, due to a short
So checking wires or wire bundles for overcurrent, is also a good diagnostic.
I have one, that has 40A and 400A full scale DC ampere readout ranges.
I use the 40A DC range for computer checks. You can grab four 5V wires on
the main power connector, and clamp the jaws around all of them, and measure
the total current flow from 5V into the motherboard. The measurement is made
via the magnetic field, and the fields of the wires add together. In about
60 seconds, I can do a quick review and see if anything is at the "burning"
level. The pins on the main connector are good for at least 6 amps each. If
four wires are connected to the same rail, that would be room for 24 amps
to flow. Due to current hogging, a typical distribution in the wires, would
be 6,5,5,4 when measured individually (you could draw 20 amps total, before the
most heavily loaded pin becomes a concern). So you cannot realistically expect
to draw equal currents in all of them. The wires are joined at the motherboard,
and the degree of equalization depends on how much copper there is to connect
them together. In at least one crappy design, some wires received a
load and certain pins tended to burn. Modern computers probably aren't close
enough to those limits, to be a concern. (Only 12V wiring now is heavily loaded,
and 12V wiring uses things like PCI Express connectors to carry the heavier
(Mine is the 380947. The keyword here is "DC amps" as not all of them do DC.
There are some clamp-on units that only do AC. I don't remember the price, but
it may have been in the $300 range. It is the most expensive instrument I have
In terms of the computer tolerance, there is real and perceived tolerance.
A switching regulator design, can accept a wide range of inputs by design.
For example, I wouldn't be surprised to find a Vcore regulator design that
could work all the way down to 6V. There is a reason they don't allow
that though. If a high end processor drew 12V @ 10A from Vcore, operating
at 6V would cause the current to climb to 20A. The "input watts" need to be
preserved. Switching regulators can be equipped with "undervoltage cutoff",
to prevent excessive current flow due to that situation. So the Vcore chip
probably shuts off the processor, if the input was only 6V. Other than that,
a switching regulator can be made to accept a relatively wide input range.
A peripheral like a disk drive, has a couple issues associated with it.
It seems to have voltage monitoring, which is relatively tight. The drive
might spin down, if the rails hit +11V instead of +12V. There seems to be a
"power bug" monitoring the rails, and indicating a local "power good" condition.
And the controller board also has protection devices for overvoltage, placed
right across the rails. If you raise the rails a bit (say 15V), the protection
device starts to burn on the disk drive controller board. The protection device
is intended to clamp shutdown transients and is not intended to do battle with
out-of-spec supplies. At least one poster, managed to figure out this little
known fact (did his own research, looked up the part number on the burned
item), and he posted details of what he learned about it.
So the computer doesn't really care your supply is at +11.5V right now.
You'll hear spindown/spinup noises from your hard drives, if you got
closer to 11V perhaps (I don't know what the exact design value is).
But the supply is likely indicating to you, that something is not right.
Which is why I like the clamp-on ammeter as the next step. if there are
no outrageous currents present, then the supply is knackered. It may be
warning you of its impeding doom. But don't take the word of the
Antec on the actual value. I didn't see any mention of voltage tolerances
in the Antec information. My $100 multimeter (for checking volts) is only
good to +/- 1.5%. You can do much better than that, going up the price
range. That was the best I could do by taking a 10 minute drive to the
I don't check amperes with the multimeter for a few reasons. Max current
on a typical home multimeter is 10 amps (fuse protected). There is the
nuisance of making up a test harness, to interrupt the wires individually
and make measurements. There is the danger the test harness could fall
apart at an inopportune moment. The clamp-on meter costs a lot of money,
but the solution is so much cleaner. I also use it for working on the
car (my alternator is currently in the process of failing), and take
measurements on the central air with it (AC range). So I get more than
just computer usage from it.
If you don't want to budget for any more test gear, just swap in another
ATX supply and see what happens.
If you could use the Antec while the system was running, it would be
interesting to see how low the +12V gets, just at the instant it loops.
If you have a multimeter with peak detection, after a few attempts you
may record a sample at the lowest point it achieves before looping.
I think my multimeter has "min" and "max" recording options on volts,
and it will record the lowest voltage value (it takes discrete samples
so there is a good chance of missing the real min or max).
Re: Low Voltage on 12V Rail
On Sun, 07 Mar 2010 00:06:11 -0600, Grinder
You didn't mention the model of HP system, but that
situation is common in the last years systems used primarily
5V power and had a low current 12V rail. I've seen systems
run many years fine like that.
On the other hand if the system uses the 4 pin, 12V
connector for CPU, it should have more regulation on the 12V
rail and capacity to result in a higher 12V reading... it
would be a sign the PSU is running beyond it's regular max
capacity... but that seems unlikely if the system is in
stock configuration including stock PSU still, versus having
adding a substantial upgrade that draws more current than it
was originally designed to support.
I'd get ahold of a multimeter to take more readings, for all
we know it could be a poor PSU tester rather than the PSU.
Probably not, but if it were it could really throw a loop at
your troubleshooting efforts.
Re: Low Voltage on 12V Rail
On 3/9/2010 4:21 PM, kony wrote:
It's an a220n.
It does have a 4-pin CPU connector, and it's in nearly a stock
configuration. A GeForce MX 440 AGP video card has added because the
onboard VGA connector has some discontinuities, and has become unreliable.
I mentioned to the owner that there *might* be a problem with the power
supply, that its rating is borderline, and that it becomes less
effective as it ages, and they opted for a replacement.
- » Newest MemTest86, MemTest86+ incompatible with Sandy, Ivy chipsets?
- — Next thread in » Computer Hardware
- » Soyo mobo doesn't shut off, but reboots after "shut off" command.
- — Previous thread in » Computer Hardware